⚙️Annotation Settings

This is the most critical configuration step.

A. Class Labels

Create the object classes that annotators will identify, for example:

• Car

• Pedestrian

• Truck

• Cyclist

• Traffic Cone

Each class includes:

• Class Name

• Color Code (for 3D box visibility)

• Super category (optional, e.g., Vehicle)

B. Annotation Type

Choose the annotation method for LiDAR scenes: ➡️ 3D Bounding Boxes (primary LiDAR annotation type)

This tool allows annotators to:

• Rotate boxes

• Adjust depth, width, and height

• Sync boxes across camera views

C. Attributes (Optional)

Attributes add metadata to each object.

Examples:

• Occlusion Level → {None, Partial, Heavy}

• Truncation → {0–25%, 25–50%, >50%}

• Motion State → {Moving, Stationary}

Supported attribute formats:

• Dropdown

• Text Input

• Button Selection

D. Additional Options

Your provided images show the following key setting:

• Enable Automatic Annotation ✔️ Enable Automatic Annotation (Shown in Image)

This allows AI-assisted box prediction for faster 3D labeling. Click Next.